POSITIONS AVAILABLE

We are currently (early 2008) searching for two talented researchers to join this project. People might be appointed either at postdoctoral level, or for people who have completed honours or master's degrees but have not yet undertaken the PhD, as "Plant ecology researcher" (HEW level 5). In either case, appointees would develop and lead their own research within the overall project, and would first-author the resulting papers. The two appointees would pursue research lines 1 and 2, as described below.

A scholarship is also available for undertaking a PhD in association with this project.

OVERALL OUTLINE OF THE PROJECT

Our overall aims are to identify the most significant dimensions of ecological strategy variation across plant species worldwide. This means understanding the tradeoffs that underpin those dimensions, and how the tradeoffs may shift along physical-geography gradients. The tissue traits and architectures of plant species are not only interesting in themselves, they also control ecosystem processes and define habitat and resources for animals and microorganisms.

Within this framework we propose three lines of research, including eleven specific questions or projects.

1. Evolutionary design of stem systems: dissecting trade-offs and correlations among a nexus of traits about woody angiosperm stems, including wood density, hydraulic conductance, hydraulic capacitance, ratio of sapwood cross-sectional area to leaf area, mechanical strength, breakage risk, and leaf-twig-size. 
2. Cost-benefit and game-theoretical models for the competitiveness of alternative ecological strategies or trait-combinations along environmental gradients, aiming ultimately towards an evolutionary-ecology vegetation model (EEVM) that has potential to play an important role in understanding the future under climate change.
3. Fuel properties of standing shoots and of litter derived from different species. Fire is really important in vegetation worldwide, so flammability is urgent to understand better.

The evolutionary ecology approach is that natural selection should be a good guide to where different trait-combinations are expected to be found in nature. Traits need to be effective in sustaining a species population, and importantly, to remain effective in face of competition from alternative ecological strategies.

Stems (line 1) and fuel (line 3) are two very significant issues for global vegetation modeling. They also intersect with each other. Global vegetation models with an evolutionary-ecology perspective (line 2) encapsulate the central challenge of plant ecology. Our ARC project 2003-2007 has brought us to promising lines of attack on each of these issues.

RESEARCH LINE 1: TRADE-OFFS AMONG TRAITS OF ANGIOSPERM STEMS

The amount of sapwood cross-section per unit leaf area, and hydraulic properties of the sapwood, are key quantities in vegetation. Soil-to-leaf hydraulic conductance per unit leaf area is needed as input to earth system models that track photosynthesis and transpiration on 10-20 min timescale resolution (e.g. land surface treatment in Katul et al 2003). The design tradeoff is that investment in sapwood subtracts from potential investment in leaf area, but leaf area that is better supplied with water can be more productive.

            Prior to the last grant cycle we had found that shoots of larger-leaved species tended to carry more leaf per twig cross-section (Pickup et al. 2005; subsequently confirmed over a wider range by Wright et al. 2006b). Also larger-leaved species tend to have lower-density sapwood (Ackerly 2004; Pickup et al. 2005; Wright et al. 2006a). Our working hypothesis was therefore that lower-density wood connoted higher hydraulic conductivity, offsetting the lower sapwood area per leaf area. But latest results have suggested little correlation between wood density and conductivity (for angiosperms, gymnosperm wood having different organization). Rather there seem to be two dimensions of variation, one of lower wood density together with larger leaf size and twig size, the other of conductivity together with canopy height. Sapwood area per leaf area was influenced separately by each dimension.

            We have also been investigating the wood-anatomy underpinnings of conductivity and wood density. Because conductivity is proportional to (vessel diameters) to the 4th power multiplied by vessel number per unit cross-section, species with similar conductivity would lie along a slope approx -4 in a log-log space of vessel number vs lumen diameter.  Actually within sites, species varied along a slope approx -2 . To lower right were species with higher conductivity due to larger but fewer vessels. Higher conductivity was achieved without increasing the fraction of cross-section occupied by vessel lumen, and so without influencing wood density in that way. Rather, the main influence on wood density was the density of the 85-98% of the cross-section that was not vessel lumen. Low tissue density outside the vessels corresponds to high water-holding capacity. The ecological value of this stem capacitance is presumed to lie in filling overnight, then delivering more water to leaves during the ensuing day, so that the transpiration supported can actually be higher than would be expected from the stem conductance (Gartner and Meinzer 2005). (Bucci et al. 2004; Meinzer et al. 2004; Santiago et al. 2004) found that higher stem capacitance was associated with less negative mid-day leaf water potentials Ψmidday, and (Goldstein et al. 1998) found it was associated with later stomatal closure. These comparisons were across relatively few species, however (6, 4, 20, and 5 respectively), so we think it important to ask across a wider range Question 1: Are lower wood densities or higher water-holding capacities consistently associated with lower Ψmidday or with a wider gap between Ψmidday and Ψpredawn?

A promising alternative working hypothesis about lower wood density is that it permits cheaper shoot extension (height growth) at the expense of a higher incidence of breakage (Falster 2006). Hence Question 2: Is lower wood density associated with more economical extension of individual twigs but also with more loss of tip twigs and rebranching from further back? The key test for this will be across species in the field, but we would also quantify mechanical strength of twigs with a Universal Testing Machine. We would develop theory for the tradeoff by expressing breakage risk in the currency of setbacks to shoot extension. A spinoff benefit of this research is about leaf lifespans, which are central to the leaf economic spectrum (Westoby et al. 2002; Wright et al. 2005a). Existing data and theory about leaf lifespan comes from leaf age sequences that occur longitudinally along shoots, but we have observed in the field that leaf loss through breakage of twigs can be substantial.

In light of the research discussed above, another promising line of attack for clarifying the nexus of stem-traits is Question 3: Across what comparisons does the pattern apply where conductivity varies independently from the fraction that is vessel lumen and from wood density? Results so far indicate it applies across species within sites, and across soil nutrient comparisons. At lower rainfall, vessel densities are lower at a given mean vessel diameter, and this does correspond to higher wood densities. It is important to investigate along latitudinal temperature gradients, and in the process to confirm the observed patterns within sites, with rainfall and with soil nutrients.

Question 4 is: Does sapwood cross-sectional area per leaf area (SA:LA) vary as predicted by a model (ms in prep?) SA:LA at both tree-base and terminal shoot levels varies considerably across species, between sites within species (Whitehead et al. 1984), and from wet to dry seasons  by shedding and adding leaf area (Eamus and Prior 2001). SA:LA variation is probably the major source of variation in whole-plant conductance per leaf area. Models for optimal SA:LA have been developed by (Buckley and Roberts 2006) and by us (in prep). A range of specific predictions arise from the two models. Optimal SA:LA is predicted to peak at intermediate levels of many variables, including height, VPD, soil texture, soil moisture and atmospheric CO2. This prediction arises because as it becomes more difficult to supply leaves with the transpiration they need, at first a greater investment in sapwood is favoured, but later there are diminishing returns and plants perform better by operating with reduced transpiration.

We believe a valuable next step would be to gather together the scattered literature observations of SA:LA worldwide into a synthetic dataset. Literature data might prove sufficient to give some direct tests of model predictions, and in any event would have the benefit of putting fresh data in a wider context, as in our "glopnet" project (Wright et al. 2006c; Wright et al. 2005a; Wright et al. 2005b; Wright et al. 2004b). Fresh data will also be collected in conjunction with the fieldwork for Q3. Because soil water supply and evaporative demand are expected to have separate effects on SA:LA, we will compare across species that show different predawn leaf water potentials within rainfall zones.

RESEARCH LINE 2: MODELS FOR WINNING STRATEGY MIXTURES IN RELATION TO GLOBAL VEGETATION

The current generation of global vegetation models (e.g. Cramer et al. 2001; Krinner et al. 2005; Sitch et al. 2003; Woodward et al. 2004) are a very substantial accomplishment.  Working largely from physiological first principles, they predict zonation of vegetation, habitat structure and organic matter production across the world's ecosystems. In the current generation, variation across plant species is represented through a set of plant functional types (PFTs), which have predetermined trait-values.  We propose to develop theory that tackles the two most important limitations of this treatment.

Question 5: Predicting the spread of values for ecological traits across coexisting strategies. The observed spread within sites is wide, e.g. typically 50-60% of the worldwide spread for traits including seed size and leaf mass per area (Westoby et al. 2002). Within-site trait variation reflects different styles of making a living, rather than adaptation to different physical environments. Strategies within a site interact game-theoretically, meaning that the success of each strategy depends on which other strategies are present.

Interactions through taller plants shading shorter plants illustrate the current state of theory, as well as being probably the most important single influence on the trait-mixture in vegetation. For heights to form a stable mixture of strategies at a point in time, the theoretical requirement is that costs increase nonlinearly with height (Iwasa et al. 1985). This prevents the tallest heights from deploying much leaf area, and the mixture is invasible by shorter strategies. From this platform, three substantial advances are now within reach.

First, the abstract Iwasa concept can be integrated with established physiological models that have realistic parameterization for rates such as stem tissue respiration, leaf photosynthesis, and tissue turnover. As proof-of-principle, a draft simulation by Falster (unpublished) in our lab is capable of predicting three key quantitities: the top-of-canopy height strategy, the total leaf area accumulated per ground area (i.e. the total shading power that prevents shorter strategies from persisting below it in the canopy), and the spread of height strategies between these upper and lower bounds.

The second advance achievable is a fast solver for mixtures that are both ecologically and evolutionarily stable in game-theoretical situations of this type. In simulations, after an approximate ecological equilibrium is reached, the strategies then evolve to different trait values, and then a new ecological equilibrium must be established, and so forth. Considerable computing time can be needed before both ecology and evolution reach quasi-equilibrium, even for a single simulation run. Adaptive dynamics theory (e.g. de Mazancourt and Dieckmann 2004; Dieckmann 1997; Doebeli and Dieckmann 2000) predicts analytically when a mixture of ecological strategies will outcompete a single strategy, and can provide mathematical tools and code for short-cutting to strategy mixtures that are stable both ecologically and evolutionarily.

Then founded on a fast solver for a single tradeoff, the interplay among multiple tradeoffs can be investigated. The Iwasa tradeoff for height is about costs and benefits at a point in time, but height growth trajectories over successional time following disturbance are also important (Falster and Westoby 2003; Falster and Westoby 2005).

Question 6: Integrate models for growth physiology, for demography and for strategy mixtures into an Evolutionary-Ecology Vegetation Model (EEVM), which makes predictions about what sorts of plant species (what constellations of traits) should make up the mixture of competitively-successful strategies in different physical environments around the world. There exist well-established models (e.g. Katul et al. 2003) that incorporate state-of-the-art physiology about soil water dynamics, soil-to-leaf conductance, stomatal behaviour and leaf photosynthesis. The aim of this question is not to improve the physiology in those treatments, but rather to integrate them into the evolutionary-ecology question what plant strategies should be part of the competitively successful mixture at a site.

Candidate traits and tradeoffs for an EEVM would include the following at least: (a) leaf economics and lifespan and herbivore defence (b) sapwood area in relation to leaf area (c) water-extracting root length in relation to leaf area (d) coordination of N allocation to carboxylation vs to electron transport and with stomatal conductance (e) allometry of height growth in relation to light competition and mortality risk (f) allocation to mycorrhizas and cluster roots (g) seed size and output in relation to size at adulthood and adult lifespan (h) leafing phenology (i) resprouting capacity versus competitive growth (j) flammabilities of litter and shoots (k) feedback of litter traits to N-mineralization.

            One key principle of an EEVM is that the strategies predicted at a site should be those that are competitively successful. This may mean strategies that overshadow others or that sequester most nutrients over time, rather than strategies with maximal growth rate. Second, the effect of evolutionarily adjusting traits to optimize outcomes will be that multiple environmental factors become equally limiting at the same time (principle of equal marginal returns). Third, a mixture of strategies will typically be predicted, through a game-theoretical approach.        

Our main motivation is intellectual interest. Such an integrated model encapsulates knowledge about mechanisms, and judgment about which processes are most influential, for the purpose of predicting plant traits around the world. It gathers together the research agenda of plant functional and evolutionary ecology, and road-tests our best current understanding. A second motivation is that an EEVM will complement existing global vegetation models (DGVMs). Where DGVMs predetermine the trait-constellations permitted, through specifying a few plant functional types with set trait-values, an EEVM will give rise to trait-combinations organically as predictions from physiological and morphological tradeoffs and competitiveness criteria, across a continuous range of possible values and combinations. 

Another important role for an EEVM is to consider what sorts of plants are predicted to be most successful under conditions that do not currently exist, especially Question 7: What constellations of traits are predicted to be competitively successful under future high CO2 conditions?  Under field CO2 enrichment to 550 ppm and over a period of years within a generation, NPP increases by about 23% (Norby et al. 2005). This is in line with predictions from the standard photosynthetic model incorporated in a range of DGVMs (Cramer et al. 2001).  Current discussion about what may happen in the future revolves around the questions whether this response might be restricted in low-nutrient environments, and whether the plants establish new equilibria of allocation and turnover (e.g. Korner 2006; Norby et al. 2005). In our view, a very important question is how plants might most effectively use this extra resource to gain competitive advantage. For example in FACE experiments at low LAI, the principal mechanism for increasing NPP has been through increased leaf deployment (Norby et al. 2005). But the greater LAI should favour strategies that deploy leaves higher than the present canopy, so at evolutionary and population-dynamical equilibrium, we might expect the ESS mixture should shift to a taller range of heights, at least as much as to a larger LAI. For another example, present-day response to CO2 enrichment typically includes carbohydrate accumulation in leaves (Long et al. 2004); but this should create selection towards more vigorous export to sinks elsewhere in the plant, for competitive growth or for nutrient acquisition.

Existing global vegetation models set out to simulate the transients, the processes by which species replacement might bring about the change from present to future, but they take the set of available plant ecological strategies as given. An EEVM will be complementary, tackling the question what plant ecological strategies should be part of the competitively-successful mixture, now or under future environments, without addressing the transients. The present proposal does not directly address questions about how natural selection might create new ecological strategies in a high-CO2 world. These questions are interesting and important, but we believe that shortlisting the traits likely to be under directional selection is the right first step.

We recognize that building a first-principles evolutionary-ecology model of the world's vegetation is an ambitious aim. We do not guarantee a comprehensive model with published code within the frame of this proposal, but we do think we have a case for embarking in that direction. Our case is (1) we have identified rather specific research targets within this overall aim (2) these specific research targets are the crucial ones for the overall aim, and the time is ripe to tackle them (3) the right collaborations have been formed with partner investigators internationally (4) while tackling the theoretical problems, we shall be in close touch with the global vegetation modeling community so that improved theoretical treatments will have excellent prospects of being implemented on real geography and in connection with detailed climate and soils data.

back to Comparative Ecology Group homepage 

(Jan 2008)